ClickHere

for

FullArticle

Circulation at the entrance of the Gulf of California from satellitealtimeter and hydrographic observations

Victor M. Godínez,1,2 E. Beier,3 M. F. Lavín,2 and J. A. Kurczyn2

Received 10 August 2009; revised 21 October 2009; accepted 28 October 2009; published 3 April 2010.

[1] The surface circulation around the entrance to the Gulf of California is described fromsatellite altimetry supported by 10 conductivity‐temperature‐depth (CTD) surveys. Thesea surface height calculated from the 1/4° World Ocean Database 2001 climatology plusthe Archiving, Validation, and Interpretation of Satellite Oceanographic altimeter data(October 1992 to January 2008) were RMS adjusted to the dynamic height calculated fromCTD data; the 27.0 kg m−3 isopycnal provided the optimum reference. In the mean, thesurface circulation shows a branch of the California Current heading landward towardthe Gulf of California entrance, where it splits into two subbranches; these are separated bya cyclonic circulation attached to the coast south of Cabo Corrientes. This feature isproduced by Sverdrup dynamics and is the first observational indication that the MexicanCoastal Current is generated locally by the wind stress curl, as previously suggested bynumerical models. The global variance of the surface circulation can be separated intoseasonal (explained variance 35%), interannual (explained variance 35%), and mesoscale(explained variance 30%) components. The seasonal signal, which shows the interplay ofthe poleward Mexican Coastal Current and the equatorward branch of the CaliforniaCurrent, can be explained by a long Rossby wave model forced by the annual wind and byradiation from the coast. The interannual component is dominated by the El Niño‐Southern Oscillation, which induces in the gulf entrance an anticyclonic (cyclonic)circulation during El Niño (La Niña); this circulation includes a poleward‐flowing branch(during El Niño) parallel to the Pacific coast of the Baja California peninsula. Themesoscale variability is caused by intense eddy activity.

Citation: Godínez, V. M., E. Beier, M. F. Lavín, and J. A. Kurczyn (2010), Circulation at the entrance of the Gulf of Californiafrom satellite altimeter and hydrographic observations, J. Geophys. Res., 115, C04007, doi:10.1029/2009JC005705.

1. Introduction

[2] In his review of the eastern tropical Pacific (ETP) cir-culation, Kessler [2006] left a question mark off the entranceto the Gulf of California (GC) to indicate the poor knowl-edge available on that area, remarking that the paucity ofdata was responsible for this situation. The area in questionis where a southward branch of the California Current (CC)and the poleward Mexican Coastal Current (MCC) meet.Recent works have started to shed some light on the circu-lation in the ETP off Mexico (ETPOM), particularly on thepoleward coastal current, which since the 1960s [Wyrtki,1967] had been considered a summer‐time extension ofthe Costa Rica Coastal Current. Kessler [2006] used his-torical data banks (hydrography and surface drifters) toshow that the connection was not evident, neither in themean nor in the seasonal time scale. This had been noticed

in numerical model outputs by Beier et al. [2003], whileZamudio et al. [2007] also used numerical models to showthat the poleward MCC is probably caused by the local windstress curl.[3] The characteristics of theMCC around Cabo Corrientes

(Figure 1) in June 2003 and June 2005 were described byLavín et al. [2006] from direct observations (conductivity‐temperature‐depth (CTD) surveys and acoustic Dopplercurrent profiler (ADCP) results). The current was 90–180 km wide and 250–400 m deep, with surface speedsbetween 0.15 and 0.3 m s−1, and transported between 2.5and 4 Sv (1 Sv = 106 m3 s−1). These authors also noticed thatmesoscale eddies affected significantly the coastal currentand appeared to transport California Current water into thecoastal zone. The presence and characteristics of mesoscaleeddies have been documented in the Gulf of Tehuantepec(from direct observations) for winter [Barton et al., 1993;Trasviña et al., 1995; Barton et al., 2009]. For summer,Trasviña and Barton [2008] used satellite‐tracked surfacedrifters and satellite altimetry to show that mesoscale eddiesdominated the circulation in and west of the Gulf ofTehuantepec, while the MCC was apparently absent. Farther

1Facultad de Ciencias Marinas, UABC, Ensenada, Mexico.2Departamento de Oceanografía Física, CICESE, Ensenada, Mexico.3Unidad La Paz, Estación La Paz, CICESE, La Paz, Mexico.

Copyright 2010 by the American Geophysical Union.0148‐0227/10/2009JC005705

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, C04007, doi:10.1029/2009JC005705, 2010

C04007 1 of 15

south, between the Costa Rica Dome and Central America,Brenes et al. [2008] found that the Costa Rica CoastalCurrent is strongly affected by the presence of mesoscaleeddies.

[4] The three most relevant time scales of the ETPOMcirculation appear to be the seasonal signal (CC and MCC),the mesoscale eddies (3–5 months), and the interannualanomalies caused by the El Niño/La Niña phenomenon. Thelatter appear to consist, during El Niño, of an intensificationof the MCC and an increase in the frequency of eddyshedding [Baumgartner and Christensen, 1985; Zamudio etal., 2001, 2007; Strub and James, 2002a, 2002b; Palaciosand Bograd, 2005; Kessler, 2006].[5] Most of the studies mentioned above are based either

on data banks with sparse data in the area, on numericalmodels, or on satellite altimetry. The latter rely on hydro-graphic data banks to establish a mean condition and use anarbitrary, fixed reference level to calculate the geopotentialanomaly and from there the geostrophic circulation. AsLavín et al. [2006] showed, the effect of salinity can be veryimportant for the circulation near the coast, and closelyspaced hydrographic stations are required to detect theMCC. While Lavín et al. [2006] used 1000 m as a referencelevel, Strub and James [2002a, 2002b] used 500 m, andKessler [2006] used 450 m; this prompts questioningwhether the same dynamics are captured by the threeapproaches and whether a reference isopycnal would pro-duce any improvements.[6] In this article we use hydrographic data from 10 crui-

ses carried out in the ETPOM to establish an optimumreference isopycnal and then analyze the 15 year series ofaltimeter data, plus World Ocean Database 2001 (WOD01)hydrography, referenced to that isopycnal. We aim to studythe three main time scales of the circulation in the ETPOM,giving some emphasis to the seasonal interplay of the CCand the MCC.

2. Data and Methods

[7] The hydrographic data used in this study come from10 oceanographic cruises in the ETPOM carried out onboardthe R/V Francisco de Ulloa; the cruises were made betweenNovember 2000 and March 2007 (Table 1 and Figure 1) aspart of the Programa Oceanográfico del Occidente deMéxico (PROCOMEX). The temperature‐salinity profiles to1000 m (or to ∼5 m above the bottom if shallower) weremeasured with a factory‐calibrated CTD (SeaBird SBE‐911plus) system, with primary and secondary sensors with asampling rate of 24 Hz. The data were processed andaveraged to 1 dbar [Godínez et al., 2005, 2006, 2007a,2007b]. Salinity (S) was calculated with the PracticalSalinity Scale 1978. The potential temperature � (°C) andthe density anomaly g� (kg m−3) were calculated accordingto the Joint Panel on Oceanographic Tables and Standards[1991].[8] High‐resolution (1/4°) gridded climatologies at stan-

dard depths (WOD01 [Boyer et al., 2005]) were obtainedfrom the National Oceanographic Data Center Web site(http://www.nodc.noaa.gov). The altimeter monthly aver-aged sea surface height anomaly (SSHA) at 1/3° of resolu-tion from October 1992 to January 2008 was produced bySegment Sol Altimétrie et Orbitographie/Developing Use ofAltimetry for Climate Studies and distributed by ArchivingValidation and Interpretation of Satellite OceanographicData (AVISO), with support from Centre National d’EtudesSpatiales (http://www.aviso.oceanobs.com).

Figure 1. Grids of conductivity‐temperature‐depth (CTD)stations during Programa Oceanográfico del Occidente deMéxico (PROCOMEX) cruises: (a) November 2000, (b) May2001, (c) November 2001, (d) May 2002, (e) November2002, (f) June 2003, (g) June 2005, (h) November 2005,(i) August 2006, and (j) March 2007.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

2 of 15

[9] Daily wind data for the period 2002–2007 wereobtained from the product called “A Cross‐CalibratedMulti‐Platform Ocean Surface Wind Velocity Product forMeteorological and Oceanographic Applications” (CCMP),which contains gridded variational analysis method oceanwind vector fields, produced from all the available micro-wave radiometer data, blended with scatterometer data(National Aeronautics and Space Administration scattererand SeaWinds on QuikSCAT/ADEOS‐II). The data wereobtained from http://podaac.jpl.nasa.gov/DATA_CATALOG/ccmpinfo.html. The horizontal resolution is 0.25° × 0.25°. Thewind stress was calculated with the drag coefficient used byTrenberth et al. [1989].

3. Results and Discussions

3.1. Choosing a Reference Level and the MeanCirculation

[10] Let’s define the “altimetry referenced sea level” hr asthe sum of the monthly SSHA altimeter observations (afterremoving the temporal mean) and the temporal mean seasurface height from the surface geopotential anomaly cal-culated from the WOD01 data, but instead of using a fixeddepth as reference in the latter calculation, let’s use a“reference isopycnal” gr. To find the appropriate referenceisopycnal, field data from the locality under study needs tobe used. Let hOBS be the sea surface height that resultsfrom the surface geopotential anomaly calculated from thehydrographic observations relative to the reference isopycnalgr. The RMS error between the two sea level estimates for agiven cruise is

ð"Þ2 ¼ 1

M

Xn¼M

n¼1

�r � �OBS�OBS

� �2; ð1Þ

where M is the number of observations (CTD casts) made inthat cruise (Table 1). The RMS error depends on the commonreference isopycnal (Figure 2), and the problem is to find theisopycnal that gives the optimum fit for all the cruises. Oncethis gr is found, hr can be considered to reflect the localgeostrophic circulation (i.e., for the area covered by thecruises) between the surface and the reference isopycnal, andwe can proceed to analyze its time series. To our knowledge,this is the first time that this approach has been used to studythe local circulation based on altimeter time series analysis.

[11] Figure 2 shows the curves " = f(gr) for all the cruises.Errors fell as deeper isopycnals were used as reference,reaching minimum values for the 27.0 kg m−3 isopycnal. Forthis isopycnal, errors varied between cruises, from a maxi-mum of 5% in November 2002 to a minimum of 2% inNovember 2000, while the average for the 10 cruises was∼3%. This is an acceptable error; therefore, we chose gr =27 kg m−3 as our local reference isopycnal. The averagedepth of the gr = 27 kg m−3 isopycnal in the 1/4° WOD01data was ∼539 ± 6 m, and the topography of the isopycnalaround its mean depth (Figure 3a) suggests that a constantreference around the 500 m level may not be a goodapproximation for this area.[12] The surface circulation of the ETPOM deduced from

hr is an improvement over previous estimates: Strub andJames [2002a] used a fixed reference level at 500 m andthe Levitus and Gelfeld [1992] data, while Kessler [2006]used 450 m and gridded expendable bathythermograph(XBT) data [Donoso et al., 1994] plus a mean T‐S curvefrom the Levitus et al. [1994] atlas. The mean surfacetopography obtained from the WOD01 data referred to the27 kg m−3 isopycnal (Figure 3b) shows a gradual variationof 10 cm from the southeast to the northwest, with acyclonic intrusion covering most of the entrance to the Gulfof California, which offshore is in overall agreement withFigure 1 of Strub and James [2002a] and Figure 2 of Kessler[2006] but contains more structure off Cabo Corrientes. TheRMS difference between this geopotential anomaly and theone calculated using 539 m as reference level (shown inFigure 3c) is 0.7 cm but reaches a maximum value of 3 cmoff Cabo Corrientes. Even if the differences were not verylarge, the use of an isopycnal as reference is more appro-priate for bulk thermodynamic balances since at these depthsthe absence of diapycnal mixing can safely be assumed.[13] The mean surface topography off Cabo Corrientes

(Figure 3b) shows that the landward intrusion of the CCsplits into two branches: one heads for the GC entrance, andthe other heads for the coast south of Cabo Corrientes.While the overall landward flow pattern of circulation isapparent in previous calculations, the separation into twobranches has not been noted before. The two branches areseparated by a sea level low (pycnocline dome), whichproduces cyclonic circulation in the 220 km adjacent to thecoast; Figure 3b suggests that a poleward coastal currentshould be present SE of Cabo Corrientes. This cycloniccirculation, which also has not been reported before (Kessler[2006] noted a different weak offshore subthermoclinedome, centered at 19°N, 109°W) is consistent with Sverdrupdynamics;

V ¼ curlð�Þ=�;

where V is northward transport, t is the wind stress, and b isthe change of the Coriolis parameter (f) with latitude. Thecalculations were made with the mean wind stress curldistribution shown in Figure 3d; the resulting circulation isshown in Figure 3e. The maximum positive value of thewind stress curl to the SE of Cabo Corrientes causes apoleward coastal current (Figure 3e), which can be identi-fied as the mean MCC. Numerical models have previouslysuggested the presence of the dome [Beier et al., 2003] andalso that the coastal positive wind stress curl could be

Table 1. List of Cruises of Programa Oceanográfico del Occi-dente de Méxicoa

Dates Year Number of CTD Casts M

7–20 Nov 2000 345–13 May 2001 587–20 Nov 2001 7120–24 May 2002 7019–30 Nov 2002 747–20 Jun 2003 1153–19 Jun 2005 1525–21 Nov 2005 1587–22 Aug 2006 1417–23 Mar 2007 148

aSee station grids in Figure 1. Abbreviation is as follows: CTD,conductivity‐temperature‐depth.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

3 of 15

responsible for the mean poleward coastal current (andillustrated the sensitivity of this result to the wind productused for the simulations) [Zamudio et al., 2007]. Theseresults (Figure 3b) improve the less detailed and more‐offshore‐oriented findings of Kessler [2002] for the meanETP circulation and provide an observations‐based physi-cal explanation for the stationary forcing of the MCC.[14] Comparisons between hr (solid lines) and hOBS (color

shading) are shown in Figure 4 for the cruises of March2007 (Figure 4a), June 2005 (Figure 4b), August 2006(Figure 4c), and November 2005 (Figure 4d). The hOBS datawere objectively mapped using a Gaussian covariancefunction with horizontal length scale of 150 km, which isabout twice the local Rossby radius of deformation [Cheltonand Schlax, 1996]. The agreement was very good, with bothfields showing comparable offshore height gradients andmaxima and minima with similar absolute and relativeheights; the RMS differences between hr and hOBS are 3.5,2.9, 4.9, and 2.1 cm during March 2007, June 2005, August2006, and November 2005, respectively. Differences arenoticeable in some features, especially during August 2006(Figure 4c) near the coast toward the southeast of thesampled area.

3.2. Seasonal Variability

[15] To separate the scales of variability of hr, we firstcalculated the temporal mean (ho) and obtained the seasonal

signal (hseason) by fitting annual and semiannual harmonicsto the 15 year hr monthly mean time series:

�rðtÞ ¼ �o þ �seasonðtÞ þ �resðtÞ ð2Þ

�seasonðtÞ ¼ �a cosð!t � �aÞ þ �s cosð2!t � �sÞ; ð3Þ

where ha and hs are the annual and semiannual amplitudes,respectively, w = 2p / 365.25 is the annual radian frequency,t is the time, and �a and �s are the phases of annual andsemiannual harmonics, respectively. The residual term hresin (2) contains the subseasonal and interannual anomalies.[16] Fitting errors of amplitudes and phases, which were

calculated as described by Beron‐Vera and Ripa [2002], werevery low (not shown). The mean ho is shown in Figure 3band was discussed above. The four parameters of the sea-sonal fitting of hr in equation (3) are shown in Figure 5. Theannual component explains, on average, 32% of the totalvariance. The amplitude of ha (Figure 5a) has maximumvalues over 10 cm on the coastal zone and then graduallydecreases offshore toward the southwest. The annual phase�a (Figure 5b), which represents the month of the year inwhich the annual surface height reaches its maximum, isparallel to the orientation of the coast. The maximum sur-face height occurs in August–September close to the coast

Figure 2. Relative error (equation (1)) dependence on the reference isopycnal. Error is between theheight from the surface geopotential anomaly from the “altimetry referenced sea level” and that calculatedfrom the PROCOMEX cruises.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

4 of 15

and in October–December in the northwestern region westof the tip of the peninsula. At the southwestern extreme ofthe region the annual component shows a ∼3 cm amplitude,with maximum values during April–May; it appears to bedisconnected from the annual variability over the coast.These results indicate that there is a strong circulationconnection on the annual time scale between the CC in theNW and the MCC in the SE; this connection was describedqualitatively by Wyrtki [1967] and Baumgartner and

Christensen [1985], but this is the first time that it hasbeen shown quantitatively.[17] The southwestward propagation of the annual signal

is part of a basin‐scale pattern described by Kessler [1990]for the thermocline, just equatorward of our region; insection 3.5 we investigate the application of his explanation(Rossby wave model) of this phenomenon for our area ofobservation. The delay of the maximum in the NW relativeto the SE is also part of a larger‐scale feature, namely, the

Figure 3. (a) Anomaly of the topography (m) of the 27.0 kg m−3 isopycnal relative to its mean depth(539±6 m), from the 1/4° WOD01 data. (b) Mean surface topography and geostrophic circulation, referredto the 27.0 isopycnal, using the same hydrographic data set. (c) Mean surface topography and geostrophiccirculation, referred to 539 m. (d) The mean wind stress curl (10−7 N m−3) from Cross‐Calibrated Multi‐Platform Ocean Surface Wind Velocity Product for Meteorological and Oceanographic Applications(CCMP) wind product. (e) Sverdrup circulation calculated from the mean wind stress curl.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

5 of 15

seasonal poleward advance of coastal sea level anomalies,from central America up to Alaska, described from coastaltide gauge data by Enfield and Allen [1980] and fromaltimeter data by Strub and James [2002a]. The origin ofthis feature remains unexplained [Kessler, 2006].[18] The semiannual component contains, on average, 3%

of the total variance and is therefore much less importantthan the annual component. The amplitude (Figure 5c) tendsto decrease from 3 cm at the coast to 0.5 cm offshore,although there is more spatial variability than in Figure 5a.The phase (Figure 5d) suggests a westward progression, asin the annual case (Figure 5b). This harmonic does not showa large‐scale spatial distribution suggesting that it couldcontribute to the NE‐SE circulation connection.

3.3. Interannual and Mesoscale Variabilities

[19] The nonseasonal residuals hres were decomposed intoa set of N uncorrelated orthogonal functions using the

empirical orthogonal functions (EOF) technique [Venegas,2001]. That is,

�res ¼ F1ðx; yÞf1ðtÞ þXNn¼2

Fnðx; yÞfnðtÞ; ð4Þ

where Fn(x, y) is the spatial pattern of the nth EOF and fn(t)is the corresponding EOF time series.[20] The first EOF mode (Figure 6), which contained 54%

of the nonseasonal residual variability, represented mostlythe El Niño/La Niña interannual variability. Figure 6bshows the high correlation (r = 0.89) between the timeseries of the first EOF mode of hres (in green) and themultivariate El Niño‐Southern Oscillation index (MEI) (inblue, http://www.cdc.noaa.gov/people/klaus.wolter/MEI/table.html). The spatial pattern (Figure 6a) shows a positive(negative) sea surface height anomaly during El Niño (LaNiña) events in the whole ETPOM. The second and third

Figure 4. Contours (cm) for the geopotential anomaly height using “altimetry referenced sea level”(solid lines) and for the cruises (color scale) made in (a) March 2007, (b) June 2005, (c) August2006, and (d) November 2005.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

6 of 15

EOF modes (not shown, but see below) explained 6.3% and4.1% of the nonseasonal variance, respectively, and theirtime series suggest that they were dominated mainly bymesoscale events, although some interannual variability isalso present. Satellite images frequently show mesoscaleeddies in the area around Cabo Corrientes, and numericalmodels suggest that they may be due to instabilities of thecoastal current [Zamudio et al., 2001, 2007], that ∼2.5eddies are generated per year, and that this number ismodulated interannually by El Niño/La Niña (with moreeddies shed during El Niño).[21] The anomaly near 15°N, 106°W–114°W at the bot-

tom edge of Figure 6a is part of the effect of El Niño/LaNiña on the northern edge of the Tehuantepec Bowl, whichwe will not discuss here. However, it is well known that ElNiño and La Niña have a strong effect on the circulation ofthe entire ETP [e.g., Strub and James, 2002a, 2002b;Kessler, 2006], especially in the coastal area. Therefore,Figure 6 is only part of a larger phenomenon, and in order toput our results into a wider perspective, we applied the sameEOF analysis to the nonseasonal residues hres for the entirenortheastern tropical Pacific (Figure 7). The first EOF(Figure 7a) explains 39% of variance of the nonseasonalanomalies, and the correlation of its time series with the

equatorial MEI is high (r = 0.84) but slightly lower thanthat of the ETPOM (r = 0.89). Therefore, this mode capturedthe interannual variability produced by El Niño/La Niñaepisodes, which start at the equatorial region (not shown)and then spreads (with values larger than 10 cm) northwarduntil ∼22°N in a ∼500 km wide band parallel to the coast.North of 22°N the influence of this mode is more restrictedto the coastal area. During the 1998 El Niño, the normalizedtimes series (Figure 7a, bottom) has a value of 1, whichmeans that for that particular event, the corresponding spa-tial distribution (Figure 7a, top) reads as actual centimetersof the upward displacement of the sea level, a consequentdownward movement of thermocline, and a considerableincrease of the sea surface temperature. This classicaldescription of El Niño events resembles Kessler’s [2006]Figure 12 (bottom), who used a smoothed SSH (runningmean of 11 months) from 1993 to 2005 TOPEX/Jason datato correlate with the same quantity at the equator and 95°W.[22] The second and third EOF modes for the entire ETP

analysis, which represent 6.5% and 5.9% of the nonseasonalvariance, respectively, are shown in Figures 7b and 7c; thetime series of these modes (Figures 7b and 7c) are mainlydominated by mesoscale events, but some interannual var-iability is also evident. The frequency spectrum of the sec-

Figure 5. Amplitudes and phases of the seasonal fit to the geopotential anomaly height using altimetryreferenced sea level: (a) annual amplitude (cm), (b) annual phase (month), (c) semiannual amplitude (cm),and (d) semiannual phase (month).

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

7 of 15

ond EOF (not shown) presents three dominant mesoscalepeaks in the frequency band of 1–2 cycles per year and tworelatively high interannual peaks in the band of 0–1 cyclesper year; these frequencies validate the numerical predic-tions of Zamudio et al. [2007]. Since mesoscale variability ischaracterized by propagating phenomena like eddies, coastaltrapped waves, and possibly long Rossby waves radiatedfrom the coast, the stationary EOF modes try to capture theirvariability in only one mode, with the consequence that thephysical interpretation of the EOF modes is not easilyunraveled.[23] We have used the second and third EOF modes to

reconstruct the corresponding nonseasonal anomalies toshow (Figure 7d) that they explain very little of the totalvariance of the ETPOM. Only in the 50 km close to thecoast of the ETPOM are these two modes of some relevance(Figure 7d), explaining 5%–9% of the variance betweenthem. We conclude that the interannual variability containedin these two EOF modes does not affect the offshore me-soscale variability of the ETPOM. In the area where thespatial variability of the second and third EOFs have highvalues (50 km off the coast) satellite altimeters do notsample well. The splitting of interannual variabilitycorresponding to traveling coastally trapped phenomenawould require coastal observations that are not presentlyavailable.[24] An inspection of the frequency spectrum of the EOF

time series from the fourth to the last shows that the dom-inant frequency corresponds to mesoscale variability.

Therefore, we can separate the nonseasonal anomalies intotwo components: the interannual and mesoscale variability.[25] Let us return to Figure 6, which represents the effect

of El Niño/La Niña on the circulation at the entrance of theGulf of California. The first analysis of the possiblemechanisms by which El Niño/La Niña could affect the Gulfof California was done by Baumgartner and Christensen[1985, p. 825], who correlated coastal sea level recordswith modes of basin‐scale oceanic and atmospheric indicesand concluded that “interannual variability in the GC isassociated with the cyclonic north equatorial circulation”with an enhanced Costa Rica Current that reached polewardwell beyond the tip of the peninsula of Baja California[Baumgartner and Christensen, 1985, Figure 10]. Satellitealtimetry has considerably improved this description, and itis now well known that positive sea level anomalies trappedin the coastal 75–150 km zone travel poleward from CentralAmerica to Alaska [Chavez et al., 1998; Strub and James,2002b; McPhaden and Yu, 1999], as surmised earlier fromcoastal sea level records [Enfield and Allen, 1980; Cheltonand Davis, 1982]. Our more detailed analysis (Figure 6),which is also based on longer time series than in previousworks, shows that the circulation induced by El Niño(La Niña) at the GC entrance is in general anticyclonic(cyclonic). While the width of the affected area in theSE is ∼300 km, it is more than double that off Cabo Corrientes(Figure 6a). At the offshore edge of the affected area there is astrong poleward current (during El Niño) that splits into twobranches at ∼21°N–22°N: one of the branches continuespoleward parallel to the peninsula (on the Pacific side), while

Figure 6. The first empirical orthogonal functions (EOF) mode for altimetry referenced sea level:(a) spatial structure and (b) time series. In Figure 6b, the MEI index is shown by the blue line.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

8 of 15

the other continues toward the GC. Unfortunately, the coastalarea, where a strong signal is known to occur (from tide gaugerecords), is not well sampled by the altimeters.

3.4. Splitting of Scales During a Cruise

[26] Using harmonic analysis and EOFs, we expressed thereferenced sea level hr in the ETPOM as the sum of threesignals: seasonal (hseason), interannual (hinter), and mesoscale(hmeso). Then equation (2) becomes

�rðtÞ ¼ �o þ �seasonðtÞ þ �interðtÞ þ �mesoðtÞ; ð5Þ

that is, the variability of hr is contained in three time scales,each of which explains part of the local variance; the dis-tribution of the variance explained by each scale is shown inFigure 8. The seasonal component (Figure 8a) dominatesalong the coast, where it explains up to 60% of the variance.Away from the coast the seasonal variability does not havean important relative weight in the total variability, reachingvalues of 10% at the southwestern region. The interannualvariability (Figure 8b) is important in the whole area, with

average values ∼30% of the local variance, but it increasestoward the southwest to up to 60%. The local mesoscalevariability (Figure 8c) is lowest close to the coast (explainedvariance 10%–20%) and increases in the offshore direction,explaining over 50% of the local variance in the westernsector.[27] Using this decomposition of the variability of hr, we

can determine the time scales involved in the characteristicsof the circulation observed in specific cruises (e.g., those inFigure 4). Overall, the three time scales have the same orderof magnitude in all the area (except for the seasonal localvariance in the southwestern sector), which in principlecomplicates the understanding of which time scale explainsthe circulation variability observed at a specific time. Forexample, Lavín et al. [2006] used hydrographic data todescribe the characteristics of the MCC during June 2003and June 2005, but they only suggested possible explana-tions for some of the observed features. The splitting of theJune 2005 hr (Figure 4b) into the three components ofvariability is shown in Figure 9. An outstanding feature ofthe seasonal component (Figure 9a) is the strong cyclonic

Figure 7. EOF decomposition of the referenced sea level anomaly for the eastern tropical Pacific:(a) first mode, (b) second mode, and (c) third mode showing spatial structure (top plots) and thecorresponding time series (bottom plots). In Figure 7a, the blue line represents the El Niño‐SouthernOscillation index (MEI). (d) Percentage of the local variance explained by modes 2 and 3.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

9 of 15

eddy south of the tip of the Baja California Peninsula, whichwe will explain below in terms of the seasonal long Rossbywave forced by the wind stress curl during June (section3.5). The poleward flow off the coast of mainland Mexico(i.e., the MCC) appears to be the sum of two components, aflow from the south over the coast and a recirculation from aflow from the north, as proposed by Lavín et al. [2006].During June 2005, the general anticyclonic circulation of theinterannual variability (Figure 9b) is very weak (one orderof magnitude smaller) compared with the seasonal and themesoscale variability (Figure 9c). The mesoscale wasdominated by cyclonic eddies, some of which alsorecirculated toward the coast, feeding water from the CC tothe MCC (also suggested by Lavín et al. [2006]). Notice that

because of the rough spatial resolution of altimetry data, thecoastal MCC was not well defined, but it can be seen in theobserved data in Figure 4b. This triple variability scalesplitting can be performed for all our 10 cruises and for anycruises made in the area while satellite altimeters are inoperation. Of course, the splitting can be performed for anydate of the altimeter record.

3.5. Rossby Wave Model of the Annual Variability

[28] As a first approximation, the spatial distribution ofamplitudes and phases of the annual cycle can be explainedin terms of long Rossby waves forced by the wind stress curland by radiation from the coast, both at the seasonal timescale. Kessler [2002, 2006] used a simple, linear, 1 1/2‐layermodel of long, nondispersive, wind‐forced Rossby waves toexplain the seasonal surface circulation in the whole ETP.As the Rossby waves move toward the west, they are dis-

Figure 8. Local variance of the referenced sea level anom-aly contained in the three time scales: (a) seasonal, (b) inter-annual, and (c) mesoscale.

Figure 9. Circulation during the June 2005 cruise (Figure 4b)split into three time scales: (a) seasonal, (b) interannual, and(c) mesoscale.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

10 of 15

sipated by friction and amplified (dampened) by coherent(out of phase) wind stress curl.[29] The equation describing the quasi‐geostrophic forc-

ing of long Rossby waves by the wind stress curl, in termsof the sea surface height (Kessler [2006] wrote it in terms ofthe thermocline depth anomaly), is

�Ro ¼ g0

gcr

Zx

xE

e�Rðx�x0 Þ

Cr Curl � x0; t � x� x0

Cr

� �=f �

� �dx0

þ e�Rðx�xE Þ

Cr �E x; t � x� xECr

� �; ð6Þ

where hRo is the sea surface height produced by the windstress curl, r is the water density, R is the damping timescale, with typical values between 6 and 9 months (Kessler[2006] and references therein), f is the Coriolis parameter,b is its meridional derivate, g′ is the reduced gravity, g is thegravitational acceleration, c is the internal gravity wavespeed, cr = bc2/f 2 is the long Rossby wave speed, and xE isthe lower limit of integration (the coast) at a given latitude.As the integration is carried westward, dx is negative. Thefirst term on the right‐hand side of equation (6) is thecontribution of the local wind stress curl to the long Rossbywaves in the interior ocean, and the second term is thewestward radiation of long Rossby waves from the coast.

[31] To evaluate the main aspects of equation (6) around20°N, we used c = 200 cm s−1 (for example, a typical valueused in the Gulf of California is 165 cm s−1 [Ripa, 1997])and R = (6 months) −1, which gives cr = 2.3 cm s−1 and ane‐folding length scale of 380 km from the coast; for example,at 380 km off the continental coast, nE will diminish by 34%,while the phase will change by 6 months. The values of theparameters are only typical; the linear solution of equation (6)can only give features of the upward and downward move-ments of the sea surface height at the seasonal time scale. Aswas pointed out byKessler [2006], the nonlinear terms are notwell understood, which makes it difficult to estimate howmuch of the variability the linear model can capture. How-ever, we can determine, in selected areas of the ETPOM,whether the solution of equation (6) consists of local windstress curl forcing of hRo and/or westward radiation from thecoast.[32] The annual variability of the wind stress curl is

shown in Figure 10 from June to September (as they weregenerated with the annual amplitude and phase, the sameplots with opposite signs apply from January to April). FromJune to September there are three areas where the windstress curl has intense absolute values, O(10−7 N m−3),highlighted by shaded areas in Figure 10: (1) around the tip ofBaja California Peninsula, with positive values up to 2.5 ×10−7 N m−3 during June; (2) in the region between the con-tinental coast and 100–200 km offshore, with maximum

Figure 10. Contours for the seasonal fit of the wind stress curl (10−7 N m−3) in (a) June, (b) July,(c) August, and (d) September.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

11 of 15

values (∼−2 × 10−7 N m−3) around Cabo Corrientes fromJuly to September; and (3) in the area located at the southwestof the region, with maximum of 0.5 × 10−7 N m−3 duringAugust. In these three regions the transfer of wind stressvorticity to the surface ocean is strong.[33] Taking the region off the tip of the peninsula with an

average value of 1.5 × 10−7 N m−3 over a length of 150 km,the first term on the right‐hand side of equation (6) predictsa local maximum value of hRo ∼ 7 cm during November–December, which is similar to the values of ha and �a inFigures 5a and 5b, respectively. In the region between thecontinental coast and 150 km offshore, the average windstress curl is ∼1.5 × 10−7 N m−3 during August, and the firstterm of the right‐hand side of equation (6) predicts that haand �a would have maximum values of ∼7 cm duringSeptember. In the area located at the extreme southwest ofthe area we can assign to the wind stress curl an averagevalue of 0.5 × 10−7 N m−3 in a length of 400 km duringMarch, which corresponds, using the first term of the right‐hand side of equation (6), to the 4 cm during April for haand �a in Figures 5a and 5b. Away from these regions thewind stress curl is null or has values that are lower by oneorder of magnitude, and the first term of the right‐hand sideof equation (6) predicts that the long Rossby wave onlypropagates westward while decaying by friction: this can beseen in Figure 5a, where there are no other places withrelative maximum amplitudes, and in Figure 5b, where thephases gradually increase westward.

[34] At the southeast extreme of the ETPOM, seasonalaltimetry data show maximum values of 4 cm in midSeptember (Figures 5a and 5b). These values cannot beattributed to the local wind curl forcing and could be due tocoastal trapped events. As a consequence, the westwardpropagation as a long Rossby wave can be explained by thesecond term of the right‐hand side of equation (6), which isvalid for the entire ETPOM coast. This term can also explainthe amplitudes of 10 cm during the middle of August northof Cabo Corrientes, which in turn cannot be explained onlyby the local wind curl forcing of the linear model. However,with monthly altimeter observations we cannot detectwhether there are seasonal Kelvin‐like signals runningpoleward trapped in the first 70 km off the coast because attheir phase speed of ∼1.65 m s−1 [Beier, 1997; Ripa, 1997],they would cross the ETPOM in ∼10 days.[35] The analytical solution (equation (6)) was evaluated

numerically using as forcing agents the annual componentof the wind stress curl (Figure 10) and the annual compo-nent of hseason at the coast (Figures 5a and 5b). Amplitudesand phases at the annual time scale of the solution are shownin Figure 11, which can be compared with the altimeterobservations of amplitudes and phases (Figures 5a and 5b).Comparison can only be qualitative because equation (6)does not include horizontal diffusion or advection; forexample, the effect of the annual variability of the CaliforniaCurrent in the interior ocean is not included. The solutionshows relative maximum amplitudes in the same areas asthe observations, i.e., south of the tip of Baja CaliforniaPeninsula, off the mainland coast, and at the southeastextreme of the sampled area. Away from these areas, thesolution shows westward propagation with a phase similarto the progression shown by the annual component of hr(Figures 5a and 5b). Near 22.5°N, 113°W the phase of thesolutions does not follow the observed phases; this couldbe due to the simplicity of the model since by notincluding lateral diffusion or advection, the solution ateach latitude is independent from the others. For example,when the coast jumps from the continental coast to the pen-insula coast at 23°N, the model phase shows a discontinuity.[36] Regarding the large‐scale connection between the

equatorward branch of the California Current and thepoleward MCC [Kessler, 2006], it appears that, exceptduring strong El Niño/La Niña conditions, the seasonalvariability is the most relevant time scale (Figure 5). Thenorthwest‐southeast connection at the annual scale (the mostimportant part of the seasonal variability) can be shown byusing hr and the Rossby wave model. The annual evolutionof the observed and modeled circulation from April to Julyis shown in Figure 12 (the opposite signs of these plots applyfor September–December). In April and May (Figures 12a–12d), the cyclonic phase of the long Rossby wave covers thewhole ETPOM. The general cyclonic circulation during May(Figures 12c and 12d) can be described as an incoming flowfrom the California Current System that reaches the southeastextreme of the region, where a branch is recirculated towardthe coast. During June (Figures 12e and 12f), the long Rossbywave has moved westward and has been dampened by fric-tion, although at the tip of peninsula the favorable wind stresscurl continues to force an intense cyclonic circulation. DuringJuly (Figures 12g and 12h) the anticyclonic phase of thelong Rossby wave starts to grow over the continental coast.

Figure 11. Modeled annual long Rossby wave characteris-tics: (a) amplitude and (b) phase.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

12 of 15

After the westward propagation from the coast, the anticy-clonic phase of the long Rossby wave will produce inOctober a spatial distribution pattern opposite to that of May(Figures 12b and 12c). The differences at the west side ofthe southern Gulf of California and the tip of the peninsula(where there is a discontinuity) could be improved if theannual Kelvin‐like signal running along the whole GC coast[Beier, 1997] was included. As mentioned before, the annualsurface dynamics calculated with the Rossby model givesonly a qualitative picture of the annual circulation, but it isencouraging that both fields, observed and modeled, showthe same southeast‐northwest connection.

4. Conclusions

[37] Using CTD data from 10 recent cruises (2000–2007)and the combination of the 1/4° climatological hydrographic

data bank of WOD01 plus altimetry observations (October1992 to January 2008), the long‐term mean surface dynamicsof the ETPOM appear to be best represented, in a root meansquare sense, when it was referred to the 27 kg m−3 iso-pycnal. Using standard statistical methods, we performed arobust decomposition of the surface variability, representedby the dynamic height and the associated geostrophic cir-culation, into three main components: mesoscale, seasonal,and interannual. The surface dynamics at the ETPOM showsintense seasonal, interannual, and mesoscale variabilities,each explaining comparable amounts of the global variance.The local variance shows that the seasonal variability isdominant in a coastal band ∼300 km wide, while mesoscalevariability accounts for most of the local variance beyondthat coastal band. The interannual variability is significantthroughout the ETPOM.

Figure 12. Evolution between April and July of (left) the annual component of the “altimetry referencedsea level” and (right) the annual long Rossby wave model.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

13 of 15

[38] The mean circulation shows a newly reportedcyclonic circulation attached to Cabo Corrientes, which canbe explained by Sverdrup dynamics, i.e., as the result of thelong‐term mean wind stress curl. This cyclonic circulationseems to be the first observational evidence that in the meanthe MCC is generated locally by the wind stress curl and istherefore disconnected from the equatorial circulation.[39] The seasonal anomalies of the sea surface height and

the associated geostrophic surface circulation can beexplained in terms of the westward propagation of longRossby waves of annual period. The superposition of thelong Rossby wave forced by the wind stress curl andthe freely propagating long Rossby wave radiated from thecoast produces, at the annual frequency, a large‐scalecommunication between the California Current System andthe tropical region. The spatial pattern is a seasonallyreversing long Rossby wave with the cyclonic phase fromMarch to July and the anticyclonic phase from October toDecember.[40] The interannual variability shows a large‐scale anti-

cyclonic (cyclonic) circulation during El Niño (La Niña).For the period of strong events, the interannual variabilitycould dominate the circulation and even change the sea-sonal‐scale spatial pattern of circulation. The equatorialinterannual variability spreads northward and affects theETPOM in a band ∼500 km from the coast. The mesoscalevariability shows that eddies account for an important partof the variance and therefore could have an important role inthe transport of properties from the coast to the interiorocean.

[41] Acknowledgments. This is a product of project “Dinámica,termodinámica y producción primaria de la Corriente Costera Mexicana”supported by CONACyT (SEP‐2003‐C02‐42941/A‐1) and CICESE.Cruises made between 2000 and 2004 were part of CONACYT projectG34601‐S. V.M.G. and J.A.K. held CONACyT scholarships. We are grate-ful for the support of the skipper and crew of the R/V Francisco de Ulloa.

ReferencesBarton, D., M. L. Argote, J. Brown, M. Kosro, M. F. Lavín, J. M. Robles,R. L. Smith, A. Trasviña, and H. S. Vélez (1993), Supersquirt: Dynamicsof the Gulf of Tehuantepec, México, Oceanography, 6(1), 23–30.

Barton, E. D., M. F. Lavín, and A. Trasviña (2009), Coastal circulation andhydrography in the Gulf of Tehuantepec, Mexico, during winter, Cont.Shelf Res., 29, 485–500, doi:10.1016/j.csr.2008.12.003.

Baumgartner, T. R., and N. Christensen (1985), Coupling of the Gulf ofCalifornia to large‐scale interannual climatic variability, J. Mar. Res.,43, 825–848, doi:10.1357/002224085788453967.

Beier, E. (1997), A numerical investigation of the annual variability in theGulf of California, J. Phys. Oceanogr., 27(5), 615–632, doi:10.1175/1520-0485(1997)027<0615:ANIOTA>2.0.CO;2.

Beier, E., M. F. Lavín, J. Gómez, V. M. Godínez, and J. García (2003), Lacorriente costera Mexicana, in GEOS, 23, 152.

Beron‐Vera, F. J., and P. Ripa (2002), Seasonal salinity balance in the Gulfof California, J. Geophys. Res. , 107(C8), 3100, doi:10.1029/2000JC000769.

Boyer, T., S. Levitus, H. Garcia, R. A. Locarnini, C. Stephens, and J. Antonov(2005), Objective analyses of annual, seasonal, and monthly temperatureand salinity for the World Ocean on a 0.25° grid, Int. J. Climatol., 25,931–945, doi:10.1002/joc.1173.

Brenes, C. L., M. F. Lavín, and A. S. Mascarenhas (2008), Geostrophic cir-culation between the Costa Rica Dome and Central America, Deep SeaRes. Part I, 55(5), 608–629, doi:10.1016/j.dsr.2008.02.005.

Chavez, F. P., P. G. Strutton, andM. J. McPhaden (1998), Biological‐physicalcoupling in the central equatorial Pacific during the onset of the 1997–98 ElNiño, Geophys. Res. Lett., 25(19), 3543–3546, doi:10.1029/98GL02729.

Chelton, D. B., and R. E. Davis (1982), Monthly mean sea‐level variabilityalong the west coast of North America, J. Phys. Oceanogr., 12(8), 757–784, doi:10.1175/1520-0485(1982)012<0757:MMSLVA>2.0.CO;2.

Chelton, D. B., and M. G. Schlax (1996), Global observations of oceanicRossby waves, Science, 272(5259), 234–238, doi:10.1126/science.272.5259.234.

Donoso, M. C., J. E. Harris, and D. B. Enfield (1994), Upper ocean thermalstructure of the eastern tropical Pacific, Tech. Rep. ERL‐450‐AOML‐36,NOAA, Silver Spring, Md.

Enfield, D. B., and J. S. Allen (1980), On the structure and dynamics ofmonthly mean sea level anomalies along the Pacific coast of North andSouth America, J. Phys. Oceanogr., 10(4), 557–578, doi:10.1175/1520-0485(1980)010<0557:OTSADO>2.0.CO;2.

Godínez, V. M., E. Beier, M. F. Lavín, J. García‐Córdova, and J. Cepeda‐Morales (2005), Datos hidrográficos frente a Cabo Corrientes y en la entradadel Golfo de California durante junio del 2005: Campaña PROCOMEX‐0506,Rep. 33921, 177 pp., Dep. deOceanogr. Fis., Cent. de Invest. Cient. y de Educ.Super. de Ensenada, Ensenada, Mexico. (Available athttp://oceanografia.cicese.mx/index.php?file=reportestec2005.)

Godínez, V.M., E. Beier,M. F. Lavín, J. García‐Córdova, and C. E. Cabrera‐Ramos (2006), Datos hidrográficos frente a Cabo Corrientes y en laentrada del Golfo de California durante noviembre del 2005: CampañaPROCOMEX‐0511, Rep. 41204, 184 pp., Dep. de Oceanogr. Fis., Cent. deInvest. Cient. y de Educ. Super. de Ensenada, Ensenada, Mexico. (Availableathttp://oceanografia.cicese.mx/index.php?file=reportestec2006.)

Godínez, V. M., E. Beier, M. F. Lavín, J. Cepeda‐Morales, J. García‐Córdova, and C. E. Cabrera‐Ramos (2007a), Datos hidrográficosfrente a Cabo Corrientes y en la entrada del Golfo de California dur-ante agosto del 2006: Campaña PROCOMEX‐0608, Rep. 60383, 165 pp.,Dep. de Oceanogr. Fis., Cent. de Invest. Cient. y de Educ. Super. de Ensenada,Ensenada,Mexico. (Available athttp://oceanografia.cicese.mx/index.php?file=reportestec2007.)

Godínez, V. M., E. Beier, M. F. Lavín, J. Cepeda‐Morales, J. García‐Córdova,and C. E. Cabrera‐Ramos (2007b), Datos hidrográficos frente a CaboCorrientes y en la entrada del Golfo de California durante marzo del2007: Campaña PROCOMEX‐0703, Rep. 60384, 183 pp., Dep. deOceanogr. Fis., Cent. de Invest. Cient. y de Educ. Super. de Ensenada,Ensenada, Mexico. (Available athttp://oceanografia.cicese.mx/index.php?file=reportestec2007.)

Joint Panel on Oceanographic Tables and Standards (1991), Processing ofOceanographic Station Data, 138 pp., U. N. Educ. Sci. and Cult. Organ.,Paris.

Kessler, W. S. (1990), Observations of long Rossby waves in the northerntropical Pacific, J. Geophys. Res., 95(C4), 5183–5217, doi:10.1029/JC095iC04p05183.

Kessler, W. S. (2002), Mean three‐dimensional circulation in the northeasttropical Pacific, J. Phys. Oceanogr., 32(9), 2457–2471.

Kessler, W. S. (2006), The circulation of the eastern tropical Pacific: A review,Prog. Oceanogr., 69(2–4), 181–217, doi:10.1016/j.pocean.2006.03.009.

Lavín, M. F., E. Beier, J. Gómez‐Valdés, V. M. Godínez, and J. García(2006), On the summer poleward coastal current off SW México, Geo-phys. Res. Lett., 33, L02601, doi:10.1029/2005GL024686.

Levitus, S., and R. Gelfeld (1992), NODC inventory of physical oceano-graphic profiles, Key Oceanogr. Rec. Doc. 18, Natl. Oceanogr. DataCent., Silver Spring, Md.

Levitus, S., R. Burgett, and T. P. Boyer (1994), World Ocean Atlas 1994,vol. 3, Salinity. NOAA Atlas NESDIS, vol. 3, NOAA, Silver Spring, Md.

McPhaden, M. J., and X. Yu (1999), Equatorial waves and the 1997–98 ElNiño, Geophys. Res. Lett. , 26(19), 2961–2964, doi:10.1029/1999GL004901.

Palacios, D. M., and S. J. Bograd (2005), A census of Tehuantepec andPapagayo eddies in the northeastern tropical Pacific, Geophys. Res. Lett.,32, L23606, doi:10.1029/2005GL024324.

Ripa, P. (1997), Toward a physical explanation of the seasonal dynamicsand thermodynamics of the Gulf of California, J. Phys. Oceanogr.,27(5), 597–614, doi:10.1175/1520-0485(1997)027<0597:TAPEOT>2.0.CO;2.

Strub, P. T., and C. James (2002a), Altimeter‐derived surface circulation inthe large‐scale NE Pacific gyres: Part 1. Seasonal variability, Prog.Oceanogr., 53(2–4), 163–183, doi:10.1016/S0079-6611(02)00029-0.

Strub, P. T., and C. James (2002b), Altimeter‐derived surface circulation inthe large‐scale NE Pacific gyres: Part 2: 1997–1998 El Nino anomalies,Prog. Oceanogr., 53(2–4), 185–214, doi:10.1016/S0079-6611(02)00030-7.

Trasviña, A., and E. D. Barton (2008), Summer circulation in the Mexicantropical Pacific, Deep Sea Res. Part I, 55(5), 587–607, doi:10.1016/j.dsr.2008.02.002.

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

14 of 15

Trasviña, A., E. D. Barton, J. Brown, H. S. Velez, P. M. Kosro, and R. L.Smith (1995), Offshore wind forcing in the Gulf of Tehuantepec, Mexico:The asymmetric circulation, J. Geophys. Res., 100(C10), 20,649–20,663.

Trenberth, K. E., G. J. Olson, and W. H. Large (1989), A global oceanwind stress climatology based on ECMWF analyses, NCAR Tech. NoteTN‐338+STR, Clim. and Global Dyn. Div., Natl. Cent. for Atmos.Res., Boulder, Colo.

Venegas, S. A. (2001), Statistical methods for signal detection in climate,Dan. Cent. for Earth Syst. Sci., Niels Bohr Inst. for Astron. Phys. andGeophys., Univ. of Copenhagen, Copenhagen.

Wyrtki, K. (1967), Circulation and water masses in the eastern equatorialPacific Ocean, Int. J. Oceanogr. Limnol., 1, 117–147.

Zamudio, L., A. P. Leonardi, S. D. Meyers, and J. J. O’Brien (2001), ENSOand eddies on the southwest coast of Mexico, Geophys. Res. Lett., 28(1),13–16, doi:10.1029/2000GL011814.

Zamudio, L., H. E. Hurlburt, E. J. Metzger, and C. E. Tilburg (2007),Tropical wave‐induced oceanic eddies at Cabo Corrientes and theMaria Islands, Mexico, J. Geophys. Res., 112, C05048, doi:10.1029/2006JC004018.

E. Beier, Unidad La Paz, Estación La Paz, CICESE, Miraflores 334,Colonia Bella Vista, La Paz, Baja California 23050, México.V. M. Godínez (corresponding author), J. A. Kurczyn, and M. F. Lavín,

Departamento de Oceanografía Física, CICESE, Km. 107, CarreteraEnsenadad‐Tijuana 3918, Zona Playitas, Ensenada, Baja California22860, México. (mxcali@cicese.mx)

GODÍNEZ ET AL.: GULF OF CALIFORNIA ENTRANCE CIRCULATION C04007C04007

15 of 15